Total Prediction Accuracy Research Articles

Design of an Improved Model for Lumpy Skin Disease Prediction Using Graph Neural Networks and Multiple Modal Data Fusion

An effective Lumpy Skin Disease prediction system mitigates economic losses and ensures animal health. Conventional approaches always fall short in terms of spatial-temporal analytics, very oft coming from a single source of data samples. To this regard, this research addresses the limitations by proposing a holistically integrated approach that puts together advanced machine learning techniques with enhanced disease prediction and management. In this paper, we provide a new framework that integrates graph neural networks for the fusion of graph-structured data with multivariables, attention mechanism, and self-supervised learning into compact feature representations. First, graph convolutional neural networks are utilized for spatial-temporal analysis. GCNs use node features, for instance, epidemiological data, and edge features in the form of connectivity data to generate node embeddings that capture complex dynamics of disease spread. It improved the prediction accuracy by up to 85%, greatly improving the spatial understanding in LSD dynamics. It basically fuses satellite image data, weather data, and livestock movement records through a multiple modal data fusion approach based on multiple head attention networks according to these embeddings. This will result in an enriched feature space in which the attention weights give insights into the relative importance of each data modality. This step improved the prediction accuracy by a further 5%, making the total prediction accuracy to be at 90%. Further fine-tuning of the model is done using self-supervised learning techniques through contrastive learning: SimCLR. This pre-paid feature representations using a large data set with unlabeled historic records on animal health data and environmental data samples. It improves model performance by up to 10% when fine-tuning the representations using LSD-specific data, thereby reducing reliance on labeled data and significantly improving the robustness and scalability of the model. This resulted in an overall accuracy of 95% for the fusion model, attained from the combination of these state-of-the-art techniques. In view of such comprehensive and diversified feature space, it showed very high resilience to adversarial attacks. Veterinary applications, in particular, support interpretability through attention mechanisms and node embeddings. Therefore, this work offers an accurate, robust, and interpretable solution to predict and manage Lumpy Skin Disease, contributing seriously to the domain of veterinary epidemiology.

Random Forest and Feature Importance Measures for Discriminating the Most Influential Environmental Factors in Predicting Cardiovascular and Respiratory Diseases.

Several studies suggest that environmental and climatic factors are linked to the risk of mortality due to cardiovascular and respiratory diseases; however, it is still unclear which are the most influential ones. This study sheds light on the potentiality of a data-driven statistical approach by providing a case study analysis. Daily admissions to the emergency room for cardiovascular and respiratory diseases are jointly analyzed with daily environmental and climatic parameter values (temperature, atmospheric pressure, relative humidity, carbon monoxide, ozone, particulate matter, and nitrogen dioxide). The Random Forest (RF) model and feature importance measure (FMI) techniques (permutation feature importance (PFI), Shapley Additive exPlanations (SHAP) feature importance, and the derivative-based importance measure (κALE)) are applied for discriminating the role of each environmental and climatic parameter. Data are pre-processed to remove trend and seasonal behavior using the Seasonal Trend Decomposition (STL) method and preliminary analyzed to avoid redundancy of information. The RF performance is encouraging, being able to predict cardiovascular and respiratory disease admissions with a mean absolute relative error of 0.04 and 0.05 cases per day, respectively. Feature importance measures discriminate parameter behaviors providing importance rankings. Indeed, only three parameters (temperature, atmospheric pressure, and carbon monoxide) were responsible for most of the total prediction accuracy. Data-driven and statistical tools, like the feature importance measure, are promising for discriminating the role of environmental and climatic factors in predicting the risk related to cardiovascular and respiratory diseases. Our results reveal the potential of employing these tools in public health policy applications for the development of early warning systems that address health risks associated with climate change, and improving disease prevention strategies.

Identification of Mixtures of Two Types of Body Fluids Using the Multiplex Methylation System and Random Forest Models.

Body fluid mixtures are complex biological samples that frequently occur in crime scenes, and can provide important clues for criminal case analysis. DNA methylation assay has been applied in the identification of human body fluids, and has exhibited excellent performance in predicting single-source body fluids. The present study aims to develop a methylation SNaPshot multiplex system for body fluid identification, and accurately predict the mixture samples. In addition, the value of DNA methylation in the prediction of body fluid mixtures was further explored. In the present study, 420 samples of body fluid mixtures and 250 samples of single body fluids were tested using an optimized multiplex methylation system. Each kind of body fluid sample presented the specific methylation profiles of the 10 markers. Significant differences in methylation levels were observed between the mixtures and single body fluids. For all kinds of mixtures, the Spearman's correlation analysis revealed a significantly strong correlation between the methylation levels and component proportions (1:20, 1:10, 1:5, 1:1, 5:1, 10:1 and 20:1). Two random forest classification models were trained for the prediction of mixture types and the prediction of the mixture proportion of 2 components, based on the methylation levels of 10 markers. For the mixture prediction, Model-1 presented outstanding prediction accuracy, which reached up to 99.3% in 427 training samples, and had a remarkable accuracy of 100% in 243 independent test samples. For the mixture proportion prediction, Model-2 demonstrated an excellent accuracy of 98.8% in 252 training samples, and 98.2% in 168 independent test samples. The total prediction accuracy reached 99.3% for body fluid mixtures and 98.6% for the mixture proportions. These results indicate the excellent capability and powerful value of the multiplex methylation system in the identification of forensic body fluid mixtures.

Discrimination of piled-up neutron-gamma pulses using charge comparison method and neural network for CLYC detectors

To realize neutron-gamma discrimination of piled-up pulses in high counting rate circumstances, the discrimination performance of the charge comparison method and three neural network models to actual piled-up pulses was studied in this paper. The neural network models, including a residual neural network (ResNet), a convolutional neural network (CNN) and a fully connected neural network (FCNN), were trained and tested by seven kinds of actual experimental data sets, which included the background signals (bg), non-piled-up neutron (n) and gamma (g) pulses as well as piled-up neutron + neutron (n + n), neutron + gamma (n + g), gamma + neutron (g + n), and gamma + gamma (g + g) pulses. The labels of the seven data sets were provided after discriminating by the charge comparison method. The results showed that the charge comparison method could be applied to discriminate piled-up neutron-gamma pulses with the Figure of Merit (FoM) value of 1.0. The integration length has been optimized to be different for piled-up and non-piled-up pulses, but follow the same criterion for neutron and gamma discrimination. The FoM values are worse than that of non-piled-up signals because the baseline fluctuation is very large under high counting rate conditions, which gives lots of interference to pulse shape discrimination. The ResNet model has the highest total prediction accuracy (93.85%) for the seven signal types, and through the comparison and analysis of the inconsistent events discriminated by the charge comparison method and the ResNet model, it is concluded that the discrimination result of the ResNet model is more accurate. These results indicate that both the charge comparison method and residual neural network can be used for complicated n/γ discrimination under high counting rate conditions, and the residual neural network works better.

Human Activity Detection Using Smart Wearable Sensing Devices with Feed Forward Neural Networks and PSO

Hospitals must continually monitor their patients’ actions to lower the chance of accidents, such as patient falls and slides. Human behavior is difficult to track due to the complexity of human activities and the unpredictable nature of their conduct. As a result, creating a static link that is used to influence human behavior is challenging, since it is hard to forecast how individuals will think or act in response to a certain event. Mobility tracking depends on intelligent monitoring systems that apply artificial intelligence (AI) applications referred to as “categories”. Because motion sensors, such as gyroscopes and accelerometers, output unconnected data that lack labels, event detection is a vital task. The fall feature parameters of tridimensional accelerometers and gyroscope sensors are presented and used, and the classification technique is based on distinguishing characteristics. This study focuses on the age-old problem of tracking turbulence in motion to improve detection precision. We trained the model, considering that detection accuracy is limited by factors such as the subject’s mass, velocity, and gait style. This is performed by employing an experimental dataset. When we used the sophisticated technique of particle swarm optimization (PSO) in combination with a four-stage forward neural network (4SFNN) to forecast four different types of turbulent motion, we observed that the total prediction accuracy was 98.615% accurate.

Simulation of Forest Distribution in the Qilian Mountains of China with a Terrain-based Logistic Regression Model

Abstract Predicting vegetation distribution strengthens ecosystem management, protection, and restoration in arid and degraded areas. However, data quality and incomplete data coverage limit prediction accuracy for Picea crassifolia Kom. (Qinghai spruce) forest in the Qilian Mountains of China. Here, we used a logistic regression model combined with high-resolution vegetation distribution data for different sampling scales and digital elevation models (DEMs) to determine the potential distribution of P. crassifolia forest in the Dayekou catchment in the Qilian Mountains. We found that the model with the best simulation accuracy was based on data with a DEM scale of 30 m and a sampling accuracy of 90 m (Nagelkerke’s R2 = 0.48 and total prediction accuracy = 83.89%). The main factors affecting the distribution of P. crassifolia forest were elevation and potential solar radiation. We conclude that it is feasible to calculate the distribution of arid mountain forests based on terrain and that terrain data at 30 m spatial resolution can fully support the simulation of P. crassifolia forest distribution.

Rapid method for yearly LULC classification using Random Forest and incorporating time-series NDVI and topography: a case study of Thanh Hoa province, Vietnam

Land-use and land-cover (LULC) mapping in the complex area is a challenging task due to the mixed vegetation patterns, and rough mountains with fast-flowing rivers. In Vietnam, LULC update is not frequently. In this study, we applied a supervised machine learning (Random forest—RF) approach to mapping LULC in Thanh Hoa province, Vietnam from 2011 to 2015 utilizing multi-temporal Normalized Difference Vegetation Index (NDVI) data from MODIS, combined with topographic features. Random forest classification (RFC) reached a total prediction accuracy of 91% and Kappa coefficient (K) of 0.89 across eight LULCs. Besides, the results showed that the features extracted from time-series NDVI comprising the mean of yearly NDVI, the sum of NDVI, and the topography were the important variables controlling the LULC classification. For similar studies on the distribution of LULC, the method proposed in this study could be helpful.

Clinical significance of tumor chemosensitivity assay in patients with head and neck cancer

Objective:To evaluate the clinical significance of OncoDrug-Seq™ tumor gene detection in patients with head and neck malignancies. Methods:A retrospective analysis of 338 patients with locally advanced or unresectable head and neck malignancies admitted to the Department of Otorhinolaryngology Head and Neck Surgery, Chinese PLA General Hospital from April 2011 to February 2021. Among them, 301 patients were pathologically diagnosed as squamous cell carcinoma, 37 cases were non-squamous cell carcinoma. All patients underwent OncoDrug-Seq™ gene detection, combined with the test results and the treatment response after induction chemotherapy to evaluate whether to change the chemotherapy regimen, and to calculate the accuracy of the gene detection and the survival rate of the patient. Results:Among 301 patients with head and neck squamous cell carcinoma, the results of the drug sensitivity test were compared with the actual clinical response, the true positive rate（TP）, true negative rate（TN）, positive predictive value（PPV）, negative predictive value（NPV） and total predictive accuracy rates were 85.37%, 65.45%, 91.70%, 50.00% and 81.73%, respectively. For patients who changed chemotherapy regimen（TN+FP） and did not change（TP+FN）, the 5-year progression-free survival（PFS） was 63.45% and 80.58%（P<0.05）, respectively, the 5-year overall survival（OS） was 54.18% and 84.74%（P<0.05）. Among 37 patients with non-squamous cell carcinoma, the TP, TN, PPV, NPV and total predictive accuracy rates were 88.46%, 72.73%, 88.46%, 72.73%, and 83.78%, respectively. The 5-year PFS of the patients who changed（TN+FP） and did not change the treatment regimen（TP+FN） were 68.57% and 56.00%, and the 5-year OS was 69.26% and 48.72%, the difference was not statistically significant. Conclusion:OncoDrug-Seq™ testing has certain significance in guiding clinicians to formulate personalized chemotherapy regimens. Timely replacement of the treatment plan can enable the patient to obtain a more ideal curative effect. Compared with patients with squamous cell carcinoma, patients with non-squamous cell carcinoma can obtain greater survival benefit from this testing. However, there are still some deviations in the accuracy of the test results compared with the actual clinical situation. Therefore, the decision should be made based on the test results in combination with the clinical reality.

A Novel Multicategory Defect Detection Method Based on the Convolutional Neural Network Method for TFT-LCD Panels

Defects on thin film transistor liquid crystal display (TFT-LCD) panel could be divided into either macro- or microdefects, depending on if they are easy to be detected by the naked eye or not. There have been abundant studies discussing the identification of macrodefects but very few on microones. This study proposed a multicategory classification model using a convolutional neural network model to work with automatic optical inspection (AOI) for identifying defective pixels on the TFT-LCD panel. Since the number of nondefective pixels outnumbered the defective ones, there exists a very serious class-imbalanced problem. To deal with that, this study designed a special training strategy that worked with data augmentation to increase the effectiveness of the proposed model. Actual panel images provided by a mobile manufacturer in Taiwan are used to demonstrate the efficiency and effectiveness of the proposed approach. After validation, the model constructed by this study had 98.9% total prediction accuracy and excellent specificity and sensitivity. The model could finish the detection and classification process automatically to replace the human inspection.

Analysis on route information failure in IP core networks by NFV-based test environment

Stable and high-quality Internet connectivity is mandatory for 5G mobile networks. However, the pandemic of COVID-19 has forced global and large-scale staying at home and telecommuting in many countries. The increasing traffic has induced more pressure on networks, devices and cloud data centers. It becomes an essential task for network operators to enable their ability to automatically and rapidly detect network and device failures. We propose a highly practical method based on highly practical technology. Our method has a high generalization ability that can efficiently extract features from large-scale unstructured data and ensure high accuracy prediction. First, 997 useful features are extracted from 28GB-per-day network logs. Then, a differential approach is employed to preprocess the extracted features so as to highlight the differences between normal and abnormal states. Third, those features are refined based on the feature importance we calculated. According to our experiment, the proposed feature extraction and refinement method can reduce computation without degrading the performance. Among the five types of failures, we achieve a 100% recall rate in four types and the rest can also reach 71%. Overall, the total average prediction accuracy of the proposed method is 94%.

Student Performance Predictions Using Knowledge Discovery Database and Data Mining, DPU Students Records as Sample

This article presents the results of connecting an educational data mining techniques to the academic performance of students. Three classification models (Decision Tree, Random Forest and Deep Learning) have been developed to analyze data sets and predict the performance of students. The projected submission of the three classificatory was calculated and matched. The academic history and data of the students from the Office of the Registrar were used to train the models. Our analysis aims to evaluate the results of students using various variables such as the student's grade. Data from (221) students with (9) different attributes were used. The results of this study are very important, provide a better understanding of student success assessments and stress the importance of data mining in education. The main purpose of this study is to show the student successful forecast using data mining techniques to improve academic programs. The results of this research indicate that the Decision Tree classifier overtakes two other classifiers by achieving a total prediction accuracy of 97%.

Suicidal ideation predicted by changes experienced from pre-treatment to 3-month postdischarge from residential substance use disorder treatment

PurposeIndividuals with substance use disorders (SUD) are at an elevated risk for suicide. Abstinence and drug-related treatment outcomes remain integral to SUD treatment, but recovery incorporates more than just the absence of substance use or mental illness and including positive mental health indices in assessment and treatment of suicidality is needed. AimsThe current study investigates the role of traditional indicators of recovery, as well as positive psychology constructs, in predicting suicidal ideation following residential SUD treatment. MethodThe study utilized a longitudinal design with baseline and 3-month postdischarge follow-up assessments of 791 individuals who attended residential SUD treatment in Australia. ResultsRates of suicidal ideation decreased from baseline to follow-up, and the magnitude of change in most indices was associated with suicidal ideation at follow-up assessment. In a hierarchical logistic regression, baseline suicidal ideation, as well as a reduction in psychological distress, increase in refusal self-efficacy, and increase in self-forgiveness, emerged as significant predictors of follow-up suicidal ideation. The final model correctly classified 98.8% of participants as not experiencing SI, and 8.7% of participants as experiencing SI at follow-up, resulting in a total predictive accuracy of 86.9%. ConclusionsThe results suggested that changes in traditional recovery indices may facilitate reductions in suicidality. As a whole, changes in positive psychology indices did not add to the ability to predict suicidal ideation once traditional indices had been accounted for, but this does not preclude the importance of these indices to SUD treatment and suicide prevention efforts.

Lithological classification via an improved extreme gradient boosting: A demonstration of the Chang 4+5 member, Ordos Basin, Northern China

• CRBM and PSO can be applied to improve the prediction efficiency of XGBoost. • CRBM-PSO-XGBoost is provably high-efficient for the lithology prediction. • Processing more learning samples is effective to enhance the prediction accuracy. • CRBM-PSO-XGBoost is proved to be robust even dealing with a sparse learning data. Acquiring reliable lithology information is a critical step for geological analysis since many basic jobs in early exploration have to be completed under the application of lithological materials. Lithology prediction then is always regarded as a research hotspot in geosciences. XGBoost is proved to be more powerful on pattern recognition than classic models, as it takes advantages of gradient boosting, classification tree, regularization, and other advanced machine learning techniques, thus being more potential to provide an ideal solution for lithology prediction. Nonetheless, this model is difficult to obtain optimal results due to the employment of many hyper-parameters, and will be low-efficient when dealing with many variables. Therefore, two computing techniques, continuous restricted Boltzmann machine (CRBM) and particle swarm optimization (PSO), are introduced to improve prediction performance of XGBoost. CRBM can extract fewer while more significant features from original data, and PSO will automatically optimize hyper-parameters during training process. Data used for validation is derived from tight sandstone reservoirs of member of Chang 4 + 5, western Jiyuan Oilfield, Ordos Basin, Northern China. Three experiments are designed to verify prediction capability of the proposed model. In order to highlight validation effect, two classic predictors named support vector machine (SVM) and gradient boosting decision tree (GBDT) are applied to create a contrast. The total prediction accuracy and the respective accuracy of each lithology produced by CRBM-PSO-XGBoost are all the highest in three experiments, well demonstrating the proposed model is effective to predict the lithology of tight sandstone reservoirs and has better robustness.

Holistic analysis and prediction of life cycle cost for vertical greenery systems in Singapore

Recent decades have witnessed the emerging and development of vertical greenery systems (VGS) or green walls as an alternative to enhance the quality of urban environments, especially for high-density cities. Although previous studies have calculated the life cycle cost (LCC) of VGS, the conventional method is time-consuming and requires a large amount of detailed data, which is unable and unpractical to obtain LCC of new VGS projects by collecting data of only several key variables of VGS. Hence, after computing and analyzing the LCC of 40 cases from different projects with various types, sizes, and other attributes, our study established a prediction model for the LCC of VGS using the data from forty cases of three types of VGS. Multiple linear regression (MLR), MLR with stepwise regression, and orthogonal partial least square (OPLS) models were built and compared. And the OPLS model was selected as the prediction model as it has relatively good prediction performance and can avoid the collinearity issue encountered by the other two methods. Furthermore, the combined OPLS model for three types of VGS was separated into three individual models with improved total prediction accuracy (85.4%–95.5%). In addition, key variables affecting the LCC of different types of VGS were also identified including salary, plant density, plant price, etc. Applications of LCC prediction models for VGS for various stakeholders were discussed to conclude the study.

A Comparative Study of Chest Computed Tomography Findings: 1030 Cases of Drug-Sensitive Tuberculosis versus 516 Cases of Drug-Resistant Tuberculosis

PurposeTo investigate the CT features of drug-resistant pulmonary tuberculosis (DR-PTB) and the diagnostic value of CT in DR-PTB diagnosis to provide imaging evidence for the timely detection of drug-resistant Mycobacterium tuberculosis.Materials and MethodsA total of 1546 cases of pulmonary tuberculosis (PTB) with complete clinical data, chest CT images and defined drug sensitivity testing results were consecutively enrolled; 516 cases of DR-PTB were included in the drug-resistant group, and 1030 cases of drug-sensitive pulmonary tuberculosis (DS-PTB) were included in the drug-sensitivity group. Comparative analyses of clinical symptoms and imaging findings were conducted. Univariate and logistic regression analyses were performed, a regression equation model was developed, and the receiver operating characteristic (ROC) curve was constructed.ResultsIn the univariate analysis, some features, including whole-lung involvement, multiple cavities, thick-walled cavities, collapsed lung, disseminated lesions along the bronchi, bronchiectasis, emphysema, atelectasis, calcification, proliferative lesions, encapsulated effusion, etc., were observed more frequently in the DR-PTB group than in the DS-PTB group, and the differences were statistically significant (p<0.05). Exudative lesions and pneumoconiosis were observed more frequently in the drug-sensitivity group than in the drug-resistant group (p<0.05). Logistic regression analysis indicated that whole-lung involvement, multiple cavities, thick-walled cavities, disseminated lesions along the bronchi, bronchiectasis, and emphysema were independent risk factors for DR-PTB, and exudative diseases were protective factors. The total prediction accuracy of the regression model was 80.6%, and the area under the ROC curve (AUC) was 82.6%.ConclusionChest CT manifestations of DR-PTB had certain characteristics that significantly indicated the possibility of drug resistance in tuberculosis patients, specifically when multifarious imaging findings, including multiple cavities, thick-walled cavities, disseminated lesions along the bronchi, whole-lung involvement, etc., coexisted simultaneously. These results may provide imaging evidence for timely drug resistance detection in suspected drug-resistant cases and contribute to the early diagnosis of DR-PTB.

Analysis of scale-specific factors controlling soil erodibility in southeastern China using multivariate empirical mode decomposition

Soil erodibility (K) is a key factor in predicting the erosion of soil by water. K is spatially influenced by many environmental factors, but studies of the scales of factors influencing K are rare. The objective of this study was to identify the scale-dependent relationships between K and various environmental factors using multivariate empirical mode decomposition (MEMD). K and nine environmental factors were investigated at 101 locations along a 4200-km sinuous transect in southeastern China and then they were decomposed and analyzed using MEMD. Mean K was 0.043 ± 0.008 t ha h MJ−1 mm−1 ha−1 and had a weakly moderate spatial variability. Six intrinsic mode functions (IMFs) and a residue were obtained for K and the environmental factors after decomposition. The mean scales for IMF1 to IMF6 were 123.4, 192.4, 347.0, 598.9, 1268.3 and 1559.2 km, respectively. IMF1 for each variable explained most of the spatial variability for the variable, and IMF1 and IMF2 or IMF1 and IMF3 explained half of the spatial variability of the variable. K was mainly associated with soil-particle composition and soil organic-matter content at small scales (IMF1-IMF2), with soil-particle composition and pH at moderate scales (IMF3-IMF4) and with elevation, saturated hydraulic conductivity and clay content at large scales (IMF5-IMF6). Modeled prediction functions of K at each scale of decomposition were accurate, but the total prediction accuracy at the sampling scale was slightly lower than for multiple linear regression. The MEMD method was nevertheless valuable and provided detailed information about the factors influencing K at different spatial scales.

Tool wear monitoring system in belt grinding based on image-processing techniques

Tool wear monitoring is a major concern in securing surface quality of workpieces and performance of the machining process. Most existing tool wear monitoring techniques seem to have looked at places other than the tool itself for solution or simply inapplicable in the highly automated industries. With insights from these techniques, this paper thus proposes an image-processing-based tool wear monitoring method that combines random forest classifier (RFC) and a multiple linear regression (MLR) model to detect different wear conditions and evaluate the remaining grinding ability for robotic belt grinding. Through a non-contact digital microscope capturing images of belt surfaces, the correlation between abrasive grain area and grinding belt life is established, the tree-based RFC method is applied for belt condition monitoring, and a MLR model to grinding ability evaluation. Results from training and testing verify the validity of the proposed monitoring method: the total prediction accuracy of RFC is over 90% under different grinding belt conditions, and the mean absolute percentage error of the MLR model is less than 3.39%.

Validation of marker-assisted selection in soybean breeding program for pod shattering resistance

Significant yield losses are caused by pod shattering during the harvest of domesticated crop plants. The phenotype evaluation of pod shattering requires fully grown plants, and it is a labor-intensive and time-consuming procedure. The purpose of this study was to test the applicability of previously identified KSS-SNP5 SNP marker on chromosome 16 in the marker assisted selection (MAS) system for pod shattering in soybean breeding programs. For this research, TaqMan SNP assay was successfully developed to detect the A/G allele of the KSS-SNP5 marker in the breeding materials, including two breeding stages, F2:3 populations, and yield trial stage breeding lines. The MAS prediction accuracy was calculated by comparing the genotyping data from the TaqMan assay with the phenotype data from the dry-oven method. The total MAS prediction accuracy was 92.5% and 96.2% in two F2:3 populations and breeding lines, respectively and 85% and 96% accuracy were shown in severe conditions. In this study, we confirmed that the KSS-SNP5 TaqMan probe tightly linked to pod shattering is effective to select a resistant line. Based on these results, we propose that the marker tightly linked to pod shattering can be of valuable use in soybean breeding program for the development of pod shattering-resistant cultivars.

A data-efficient self-supervised deep learning model for design and characterization of nanophotonic structures

With its tremendous success in many machine learning and pattern recognition tasks, deep learning, as one type of data-driven models, has also led to many breakthroughs in other disciplines including physics, chemistry and material science. Nevertheless, the supremacy of deep learning over conventional optimization approaches heavily depends on the huge amount of data collected in advance to train the model, which is a common bottleneck of such a data-driven technique. In this work, we present a comprehensive deep learning model for the design and characterization of nanophotonic structures, where a self-supervised learning mechanism is introduced to alleviate the burden of data acquisition. Taking reflective metasurfaces as an example, we demonstrate that the self-supervised deep learning model can effectively utilize randomly generated unlabeled data during training, with the total test loss and prediction accuracy improved by about 15% compared with the fully supervised counterpart. The proposed self-supervised learning scheme provides an efficient solution for deep learning models in some physics-related tasks where labeled data are limited or expensive to collect.

Targeted Metabolomic Analysis of Serum Fatty Acids for the Prediction of Autoimmune Diseases.

Autoimmune diseases (ADs) are rapidly increasing worldwide and accumulating data support a key role of disrupted metabolism in ADs. This study aimed to identify an improved combination of Total Fatty Acids (TFAs) biomarkers as a predictive factor for the presence of autoimmune diseases. A retrospective nested case-control study was conducted in 403 individuals. In the case group, 240 patients diagnosed with rheumatoid arthritis, thyroid disease, multiple sclerosis, vitiligo, psoriasis, inflammatory bowel disease, and other AD were included and compared to 163 healthy individuals. Targeted metabolomic analysis of serum TFAs was performed using GC-MS, and 28 variables were used as input for the predictive models. The primary analysis identified 12 variables that were statistically significantly different between the two groups, and metabolite-metabolite correlation analysis revealed 653 significant correlation coefficients with 90% level of significance (p < 0.05). Three predictive models were developed, namely (a) a logistic regression based on Principal Component Analysis (PCA), (b) a straightforward logistic regression model and (c) an Artificial Neural Network (ANN) model. PCA and straightforward logistic regression analysis, indicated reasonably well adequacy (74.7 and 78.9%, respectively). For the ANN, a model using two hidden layers and 11 variables was developed, resulting in 76.2% total predictive accuracy. The models identified important biomarkers: lauric acid (C12:0), myristic acid (C14:0), stearic acid (C18:0), lignoceric acid (C24:0), palmitic acid (C16:0) and heptadecanoic acid (C17:0) among saturated fatty acids, Cis-10-pentadecanoic acid (C15:1), Cis-11-eicosenoic acid (C20:1n9), and erucic acid (C22:1n9) among monounsaturated fatty acids and the Gamma-linolenic acid (C18:3n6) polyunsaturated fatty acid. The metabolic pathways of the candidate biomarkers are discussed in relation to ADs. The findings indicate that the metabolic profile of serum TFAs is associated with the presence of ADs and can be an adjunct tool for the early diagnosis of ADs.